This Week’s Top 5 Stories in AI

Jeff Bezos: Why Space Could be the Future of AI Data Centres

The worldwide struggle to power AI is now pushing technology companies to look beyond Earth. 

Jeff Bezos, Founder and Executive Chair of Amazon, tells an audience in Turin that gigawatt-scale data centres will be built in space within the next 10 to 20 years – and that they’ll eventually outperform anything we can build on the ground.

Is it possible that data centres in space could be the future of powering AI demand? Jeff explains how it could be.

Speaking at Italian Tech Week, Jeff lays out his vision during a fireside chat with John Elkann, who chairs both Ferrari and Stellantis.

“One of the things that’s going to happen next – is we’re going to start building these giant gigawatt data centres in space,” Jeff says.

To put that in perspective, a gigawatt is one billion watts of power, roughly what a large nuclear power plant could produce.

How Google Will Elevate the Olympic Games Experience With AI

Taking the Olympics’ relationship with AI up another level, Google is partnering with the Los Angeles 2028 Olympic and Paralympic Games organising committee (LA28), Team USA and NBCUniversal to integrate its AI and cloud computing infrastructure across the sporting event. 

Google will serve as a founding partner and official cloud provider for the games, which will be the first Olympic and Paralympic events held in the US in 26 years.

The partnership will see Google deploy its Gemini large language model (LLM), Google Cloud infrastructure and search technologies.

Altogether, these tools will support operations spanning athlete training analysis, fan information retrieval and workforce management for 70,000 volunteers and staff members. 

Fans will access these AI features through Google Search on their phones and computers, with the technology integrated into NBCUniversal’s broadcasting to help viewers find information about events, athletes and results while watching coverage.

How AMD is Challenging Nvidia With a 6GW OpenAI Chip Pact

The battle for computing power has become the defining constraint in AI development – without one, there is not the other.

As companies across the world innovate to build ever-larger models, access to the specialised chips that train them has turned into a strategic priority.

Now AMD, the semiconductor manufacturer, has struck a partnership with OpenAI that will see the ChatGPT maker deploy 6 GW of AMD graphics processing units (GPUs) across multiple chip generations. 

“We are thrilled to partner with OpenAI to deliver AI compute at massive scale,” says Dr Lisa Su, Chair and CEO of AMD.

“This partnership brings the best of AMD and OpenAI together to create a true win-win enabling the world’s most ambitious AI buildout and advancing the entire AI ecosystem.”

The first 1GW of AMD’s Instinct MI450 Series GPUs will arrive in the second half of 2026, with deployments scaling up from there.

The deal is a big decision for OpenAI, which has relied almost entirely on Nvidia chips to train the models behind ChatGPT and its other AI products. 

How Will Perplexity’s Comet AI Impact Internet Browsing?

Perplexity is positioning its new product, Comet, as a fundamental reimagining of internet browsing, powered by AI.

Perplexity proposes that the current model is broken, having transformed the web into a transactional space at the expense of curiosity and exploration.

Comet is presented as the solution, an AI-powered browser designed to address what Perplexity views as the unmet needs of millions of internet users.

The feeling that the web’s core principles of discovery have been eroded by advertising and low-quality content is a sentiment gaining traction.

Perplexity’s marketing for Comet directly targets this nostalgia, arguing “the internet has stifled our curiosity” and has become a “digital yellow pages, where every path leads to a checkout button.”

According to Perplexity, early data indicate a major change in user behaviour, claiming that individuals who downloaded Comet increased their question-asking by six to 18 times on their first day of use.

This metric could suggest the browser fosters a more exploratory user experience, reminiscent of the internet’s earlier days.

OpenAI Targets Enterprise With App Integrations Partnerships

OpenAI has revealed a series of enterprise-focused partnerships at its 2025 DevDay conference, marking a transition from consumer momentum to business adoption. 

The company announced collaborations spanning application integrations, product development partnerships and infrastructure deals including a 6 gigawatt agreement with AMD as it seeks to justify spending that has pushed annual losses to US$8bn.

CEO Sam Altman said at a press conference following his keynote that the company would pursue enterprise clients with renewed focus. “You should expect a huge focus from us on really leaning into enterprise,” he said.

OpenAI has launched a way for apps to plug into ChatGPT allowing users to ask questions or perform tasks within third-party applications. 

The company released its Apps SDK in preview for developers, built on the Model Context Protocol, with initial partners including Booking.com, Canva, Coursera, Figma, Expedia, Spotify and Zillow. 

The company will add DoorDash, Instacart, Uber and AllTrails later this year.